Tentative update to use transformers pipeline

To use transformers pipeline with `trust_remote_code=True`:
```
from transformers import pipeline
nlp = pipeline("text-generation", "JonusNattapong/AiDaeng-Thai-RoPE", trust_remote_code=True)
text = "สวัสดีครับ"
response = nlp(text, max_new_tokens=50, temperature=0.7)[0]["generated_text"]
print(response)
```

config.json

@@ -1,20 +1,26 @@
-{
-  "_name_or_path": "./models/thai_transformer_rope",
-  "architectures": [
-    "ThaiTransformerModel"
-  ],
-  "block_size": 2048,
-  "confidence_threshold": 0.7,
-  "conv_kernel_size": 7,
-  "dropout": 0.1,
-  "dtype": "float32",
-  "max_position_embeddings": 2048,
-  "max_reasoning_effort": 1.0,
-  "model_type": "thai_transformer",
-  "n_embd": 384,
-  "n_head": 6,
-  "n_layer": 6,
-  "torch_dtype": "float32",
-  "transformers_version": "4.36.0",
-  "vocab_size": 44216
-}
+{
+  "_name_or_path": "./models/thai_transformer_rope",
+  "architectures": [
+    "ThaiTransformerModel"
+  ],
+  "auto_map": {
+    "AutoConfig": "hf_model.ThaiTransformerConfig",
+    "AutoModelForCausalLM": "hf_model.ThaiTransformerModel"
+  },
+  "block_size": 2048,
+  "confidence_threshold": 0.7,
+  "conv_kernel_size": 7,
+  "dropout": 0.1,
+  "dtype": "float32",
+  "max_position_embeddings": 2048,
+  "max_reasoning_effort": 1.0,
+  "model_type": "thai_transformer",
+  "n_embd": 384,
+  "n_head": 6,
+  "n_layer": 6,
+  "num_hidden_layers": 6,
+  "tokenizer_class": "PreTrainedTokenizerFast",
+  "torch_dtype": "float32",
+  "transformers_version": "4.36.0",
+  "vocab_size": 44216
+}